[sdiy] Is everything digital?

[phillip m gallo]

The implementation of switched digital processes to impact an analog signal
chain does not in my opinion, indicate that Crow is off the mark with his
characterization that:

"discrete time domain = digital  -    continuous time domain = analog".

While these names really are nothing more than "handles" they are generally
"accepted" as indicating precisely the above.

Referring to mixed signal implementation such as Switched Capacitor Filters;
As pointed out "There is actually two independent (orthogonal) things
happening:".  This doesn't change the nature of the two co-implemented
processes.

Switched Capacitors perform digitally to impact an otherwise analog process.
Not different from a switch implemented anywhere in an analog process (like
a selector/mux, or gated path (ON/OFF) switch).  Were the capacitors
"variable" then the mechanism would be analog but switched capacitors are
either In or Out of circuit.  The analog process they are impact integrates
their effect in the analog domain.

A comparator is indeed a digital device, a single bit Analog to Digital
converting device. That the period resulting from this conversion may or may
not be periodic (depending upon the input analog signal) the output has only
2 numerical states.

A Sample and Hold is an analog device moderated by a digital switching
mechanism and is in fact just a special case of the switched capacitor
filter, just clocked much slower.

A keyboard hosted S+H opens a sample gate based up a "pulse" derived from
the keypress activation.  At this time what ever analog level is present
attempts to charge the capacitor during the sample gates "ON" time.  The key
"un-press" closes the sample gate and retains the maximum charge achieved
within the limits of the circuit implementations capability.

A clocked sample an hold is not different just periodic even more like a
switch capacitor filter implementation. As an example, a BBD uses a digital
mechanism to impact an analog signal chain (a complimentary array of
"Sample & Hold stages). Either the feed switch is on or the pass switch is
on, (even should both switches occur "ON" the mechanism is then 2 bit
"digital".

regards,
p



You can have discrete time domain in pure analogue too. There's bunches of
switch capacitor filters out there in real life to prove me right.

There is actually two independent (ortogonal) things happening:

* Continous vs. Discrete time scale
* Continous vs. Quantified signal level scale

In traditional "analogue" you have continous time and signal level.
In traditional "digital" you have discrete time and signal level.

Put a signal through a comparator, a schmitt trigger or quantizer and you
have
continous time scale but quantized signal level. Not quite "digital".

Put a signal through a sample and hold stage, and you have continous signal
level scale, but discrete time scale. Not quite "digital".

Infact, you can have _both_ discrete time and signal level scale, without
really be "digital". In "digital" we transport signals in binary (or on odd
occasions other modulations forms like trinary), but many times we define
the
signal as being one or more of the transporting bits, and for continous
signals
(sampled) we often use Pulse Code Modulation (PCM).

I strongly recommend the reading of Shannons article on digital telephony
from
1948. Briliant motivation for the concept. In the Appendix the now called
"Nyquist theormem" is being prooved, but with a error in the proof I might
add.

Cheers,
Magnus


--
No virus found in this incoming message.
Checked by AVG Anti-Virus.
Version: 7.0.308 / Virus Database: 266.11.10 - Release Date: 5/13/05